Avalon Batch 2 is due to start shipping 15 April.
It's not known how many BFL units will ship in Batch 1 or how many TH because of the power problems they're having, but they have a maximum of 5,000 chips and it sounds like they're going to have to underclock a significant number.
There's no realistic estimate for when BFL's Batch 2 (which still need 5 layers completed, then bumping, dicing, packaging etc) and bulk batch will be ready to ship, but it's highly unlikely to be before Avalon's Batch 2.
I'd be reluctant to sell an Avalon Batch 2 order when it's extremely unlikely that BFL will iron out its problems and start shipping its own Batch 2 before 15 April.
It's unlikely BFL ships a significant amount before Batch 3 of Avalon is completed. They need a miracle to get over the thermal issues they are facing. There are 100 ways it can shake out, and 95 of them involve down-clocking the hardware dramatically. |
|
|
|
Can someone from Avalon chime in? That's probably asking a lot. I'd normally think a power supply is a power supply, but this is custom stuff and it's been a long time since I built my own towers. Knowing exactly which model is best would be great to know.
Thanks!
Seriously, why did you buy it without a power supply if you don't have enough technical knowledge to spec. one out? He does have enough technical knowledge. However he is asking for ADVICE from those who have more up to date model knowledge. He knows: He needs a PSU He needs an ATX layout PSU He needs a high eff PSU He needs a 700>W PSU but he wants to know if anyone has any recommendations or known to work [should be all] when powering his $6000 PC. And he has been told the answer 5 times now. But he wants to waste Avalon's time by getting another answer from them. So to make it dead simple: Buy a top quality Gold rated PSU rated at LEAST 25% higher than the power required by the Avalon you ordered. Go to Anandtech or 100 other hardware enthusiast sites and do your research. |
|
|
|
Can someone from Avalon chime in? That's probably asking a lot. I'd normally think a power supply is a power supply, but this is custom stuff and it's been a long time since I built my own towers. Knowing exactly which model is best would be great to know.
Thanks!
Seriously, why did you buy it without a power supply if you don't have enough technical knowledge to spec. one out? |
|
|
|
They are selling their product with a 30 day ROI price point. Yes, they are selling. But not delivering. Avalon has openly said they will mine a bit for themselves. Look at that 15 Terahash added on the 28th of March. 10 x 1,5TH BFL Mini Rigs? Then a Price Dump on MT Gox on that same day from 90ies down to 70ies (over 20% drop). I say someone just made half a million dollars, and it wasn't me with my GPUs. Also, please show me a BitCoin Miner who isn't greedy. This is a game full of crooks, kids and those who have jobs and can code (pool operators). Well I mine bitcoin. And I took $800 out of my pocket to warn newbies about BFL last week. I've bid $475 to do the same next week. Am I greedy? As for Avalon, what they said was: "If people don't want to pay the price we think is fair, we will mine for ourselves." |
|
|
|
Thats a bit much for me atm. Plus im in the USA. I'm in the US as well. |
|
|
|
Basically take a good look at Avalon's actions They have sent out emails saying they are developing a mining farm and delaying the second batch along with the third batch getting a crazy price raise. There actions are bad as BFL's in every way except for the lucky 300. The point I am making is asic's or the lack of the ability to buy them is going to kill bitcoins. That is on the bitcoin program team not avalon not bfl and not asciminer. Since a 30 or 60 day ban on all workers with 30gh hash power would once again spread the network out instead of having a hand full of people or groups with most of the hash power. ^This 100% BitCoin Devs should change algo to destroy those greedy motherf*ckers. All of them. Nonsense. They aren't being greedy. They are selling their product with a 30 day ROI price point. That is a gift. It would be nice if they could find a way to make a product that could be priced closer to an enthusiast graphics card so ASICs could be more widely distributed, but that isn't Avalons fault. ASICs are a net benefit as it is harder to attack the network now. ASICMiner is troublesome in terms of their market share, but hopefully they will be responsible and start selling hardware rather than growing their farm from where it is. And ASICs hold nearly 50% of hashpower already. Good luck putting in place a change that harms their owners. |
|
|
|
Meh, flipped the fund a quarter BTC. A full BTC seemed a bit much considering it just hit $93.
Very funny. I know BFL is probably pretty close to shipping, but I love this kinda trolling.
Thanks. Any amount is appreciated. The address has 0.7364 BTC right now. Felt like a chump for not honoring my original 1BTC pledge =( Just shot you another 0.75 BTC. Troll away ! Thanks! Address is up to 1.57 right now. And Josh has been outbid for every ad slot next week at the moment.  |
|
|
|
Their were field reports of fires in white box AMD computers at the time. AMD adamantly denied it was possible, which lead to the video demonstrations that you could cause one to catch fire by removing the heat sink.
The problem was that AMD went cheap and didn't have thermal diodes in the silicon. They left it to the motherboard manufacturers to put cheaper thermocouples on the motherboards. So it turned out there were a number of manufacturing defects* that could lead to a flaming Thunderbird.
Yeah, I've heard about smoking, flaming and fires too in the white-label reseller channels (like MWave, etc.). Because the people in my company were involved in the reselling of the AMD products I have some additional inside information: 1) on-the-motherboard thermal protection was too slow acting in the most common failure mode: spring loaded heatsink mount unsnapped because it either: 1a) was never snapped in properly all the way 1b) was snapped properly but snapping-in caused cracking of the plastic socket latch 2) rampant remarking and overclocking of AMD Athlons, including defective chips destined for industrial waste disposal. AMD Malaysia was the leaky culprit here: chips that failed tests were shipped to Taiwan or Singapore for intended destruction/recycling but it fact were remarked and shipped back to Malaysia for reintroduction into the reseller channel. This was made easier by the fact that the clock selection and feature selection (Athlon XP/MP/Mobile) traces were easily accessible on the Athlon OPGA package. Additionally AMD was really tardy in investigating and prosecuting the in-the-channel diversion. In my company they dealt with 1) and 2) with simple visual inspection after dismounting the heathsink, remounting it and doing additional burn-in with memtest86. I came at the tail edge of the problem and changed the burn in to use PRIME95 from the GIMPS prime number search. The before-burn-in failure rate was apparently quite high, I don't know the numbers because it was essentially free to us, MWave covered that. After-the-burn-in failure rate was very low, lower than what we were getting later on with white-label Dells and Supermicros. Cheap ECC RAM support in Athlons was a real godsend. I still have several of those machines in the QA farm alongside Slot-1 Intel Pentiums III. I wish somebody who was really involved in this in South-East Asia would write some sort of war memoir of the whole story. I've been to our Penang sites. AMD has a factory in the same industrial park. Our security makes the airports seem hands off. All scrap goes into grinders on site with every part accounted for by the kind of security guards you really don't want to annoy. It was a very competitive time. We learned as many hard lessons as AMD did. And BFL is so far outside of the design rules learned from those lessons that I can say with complete confidence that from here, they are going to have a really bad time. |
|
|
|
Right, the Thunderbirds. Those were neat. Instant heat death.
I kinda recall that to have the AMD Athlon release the smoke one had to remove the cooler? I worked more with Intel Itanium and Itanium 2 machines and those indeed were beastly: the CPU replacement required the wrenches from the car mechanic's toolset. Their were field reports of fires in white box AMD computers at the time. AMD adamantly denied it was possible, which lead to the video demonstrations that you could cause one to catch fire by removing the heat sink. The problem was that AMD went cheap and didn't have thermal diodes in the silicon. They left it to the motherboard manufacturers to put cheaper thermocouples on the motherboards. So it turned out there were a number of manufacturing defects * that could lead to a flaming Thunderbird. So Josh, do you have thermal diodes on those die? Do you even know how to calibrate a thermal diode properly? * One of those manufacturing defects would be bad underfill.  |
|
|
|
By the way I have an auction for half a dozen of those 'b' grade boards. https://bitcointalk.org/index.php?topic=159052.0Winning bids are in the 3 BTC range today. And I will sell domestically up to a dozen boards for 7 BTC plus shipping and escrow. |
|
|
|
Grade-a which what we are talking about is 760-880 MH/s. There may also be grade-b boards after we have built the batch and we would expect 10% of boards to come out as grade-b based on previous experience. We will adjust the build numbers to account for that so that we get enough grade-a.
That 10% number is really funny. It reminds me of when I told you that 10% of my boards were sub-par and you insisted that nobody else was having problems and I must be doing something wrong. Anyway, rising bitcoin prices heal all wounds. I love you John. You are a shining light of professionalism in contrast to certain other hardware vendors in the Bitcoin world. |
|
|
|
|
If there isn't a cancelled check from Josh to a suitable, independent and registered charity within 48 hours hasn't he finally earned himself the scammer tag?
It's obvious they have known for a couple weeks they wouldn't hit their performance spec.
So Josh, where's the proof you kept your promise?
BTW, has everyone noticed what a pretty shade of yellow highlighting Josh has these days?
|
|
|
|
We are trying to nail it down. We have some boards at 1.76w, some at 2.5 and some at 4 and one at 6 Love how the power numbers are all over the place. entropy, how high are the chances of one of these thing blowing up in someone's face? It isn't that bad. If they shipped with the kind of number Josh is talking, fires are a real possibility. You may remember videos of AMD cpus catching fire back in 2002-2003 time frame. But I don't think they are stupid enough to ship. If they had working test sockets (they don't) they could sort packages, and maybe build some mini-singles to ship. But their yield loss is going to be catastrophic. And even a mini-single is going to sound like a blender with the fan speed required. And probably hash well under specification. Remember what a pentium 4 sounded like when it was working hard? That is less power then they are talking about. And Intel spent about ... well I can't say that. But we spent hundreds of engineer years getting the cooling solution right for those beasts. |
|
|
|
1@3
3 is already the lowest bid. You have to go at least .25 higher |
|
|
|
I saw the BFL ad saying it might be a scam. Worthy investment OP. Look at my sig lol Now that they have a working protype I guess you better change that sig. Why? They still haven't shipped. When people who paid 9 months ago are getting their gear, AND it works as promised, it's time to change that signature. |
|
|
|
Entropy. You are a moron. Yes, they missed the power targets so far, but they are shipping, and the asics work.
God you are a troll.
Shipping? Really? Do you have one? What's your tracking number? Call me when you can order from them and get a tracking number the same day. BTW: Do you understand that the power miss they are talking about probably means they cannot use the board design they have shown because the power density will be too high to cool properly? It's not a small issue. There is a reason Avalon boards have their chips so widely separated. |
|
|
|
How did the production projection go down from 500 a day to 300 units per day?
Regarding whether we are a scam or not, the several thousand shipped units can speak for itself. We've used the experience gained in mass production
of our singles in our SC production cycle (we've got a MyData Pick-And-Place and an Oven for high-speed and high-volume inhouse assembly). Someone
mentioned the backlog is already 5,000. On a production speed of 500 units a day for instance, 5,000 will only take 10 days. We've already prepared the
infratructue for very high volume production, since the great portion of the orders will come in after the SC is released to the public. 5,000 units is not much
compared to orders we expect to receive after SC release, so we're very ready for it.
I hope this message has addressed some of your questions.
Regards,
Nasser
What is Nasser's full name? LOL. He is calling the gear Kano photographed "high-speed and high-volume inhouse assembly". LMAO. Notice he carefully he says nothing. "Someone mentioned the backlog is already 5,000" - he isn't saying what the backlog actually is, he's just citing 'someone'. Maybe it's 50,000. He doesn't commit himself to anything. Probably they don't even know since their order system is such a disaster. "On a production speed of 500 units a day for instance, 5,000 will only take 10 days. We've already prepared the
infratructue for very high volume production, since the great portion of the orders will come in after the SC is released to the public."Again, notice that 500 / day is a for instance, not a statement of what they can do. These guys are so full of crap. Tomorrow they are going to desperately try to stem the tide of refund requests before they go bankrupt. I suggest anyone who can think for themselves consider if they want to try to get their money back before or after insolvency. |
|
|
|
|
Fascinating stuff.
I would change red is wrong, to red is hot!
58 C is high but could be just fine. What matters is Tj the temperature at the transistor junction. This needs to stay below the design temperature or you will have reliability problems in the long term. I'm not sure what TSMC's Tj specification is for their 110 nm process, but it is probably around 75 C.
You can calculate Tj from what you have there with the thermal resistance Rjc and the power being dissapated in the device. JC being junction to case. It should be possible to look up an Rjc value for the heat sink used here.
Clearly there is some room for optimization though.
|
|
|
|
|
I have abstained. I think anyone giving money to BFL is throwing it in a hole.
The exact flavor and odor of said hole is of absolutely no interest to me.
I do think it is time to banish all BFL threads to speculation. Keeping them in Custom Hardware is a joke. When they can show videos of a happy Garr and Fefox unboxing their toys and making coins with them, let them back into this part of the forum.
|
|
|
|
|
Show Posts
